Nondestructive Evaluation of Fig Fruit Maturity Using Visible / Short-Wave Near-Infrared Spectroscopy

In the present study, visible/shortwave near-infrared reflectance spectroscopy (Vis/SWNIR, 425–950 nm) was used to predict the taste index (SSC/TA) and flesh firmness of fig fruits. Besides, the efficiency of LDA and QDA classifiers in detecting ripe, semi-ripe, and unripe figs was studied based on a combination of pretreatment methods. A total of 167 fig trees were selected for the development and validation of the models. Principal component analysis (PCA) was employed to extract the principal components of the spectra. PLS performance and common spectral data pretreatment methods were evaluated using the residual prediction deviation (RPD), predictive correlation coefficient (rp), and root mean square error of prediction (RMSEP). Moreover, the efficiency of the classifiers and pretreatment methods was evaluated using the mean overall accuracy (%) of the testing samples. The highest mean value of RPDs based on the combined pretreatment method of MA + de-trending was 1.79 for flesh firmness (RMSEP = 1.64, rp = 0.845) and 0.89 for the taste index (RMSEP = 10.09, rp = 0.215).LDA and QDA classifiers had an overall accuracy of 93.33 percent (in no-pretreatment spectral data).